loom-sdk

The official Python client for Loom — the memory layer for research intelligence, by MetaCognition.

Give your research agents a persistent, temporal memory: add papers, notes, and data, then recall exactly what matters — with provenance — across sessions.

Install

pip install loom-sdk

Quickstart

Get an API key from the dashboard, then:

from loom_sdk import Loom

mem = Loom(
    org_id="lab",
    user_id="alice",
    base_url="https://api.loom.getmetacognition.com",
    api_key="sk_live_...",
)

# add a paper to memory
mem.ingest_paper(
    "10.1/astro",
    title="Astrocytic calcium signalling",
    text="astrocyte calcium gliotransmission ...",
    oa=True,
)

# search it back — each result carries its source
for span in mem.recall("calcium signalling", budget_tokens=8000):
    print(span["text"], "->", span["provenance"]["paper_doi"])

What you can do

mem.ingest_document("paper.pdf", open("paper.pdf", "rb").read())  # PDF/Excel/Word/CSV/MD/text
mem.put_analysis("10.1/astro", analysis_markdown)                 # extract claims
mem.recall_papers("astrocyte calcium")                            # find papers
mem.graph.link("session_1", used_papers=["10.1/astro"])           # link a session
mem.insights.directions()                                         # novel directions
mem.account.get()                                                 # account + usage

The client is a thin wrapper over the Loom HTTP API — every method maps to one call.

Documentation

Full docs: https://loom.getmetacognition.com

License

MIT